If you’ve kept up with OpenAI and ChatGPT in recent years, you are likely aware of the significant growth the company has experienced and the advancements made by the AI. However, this merely marks the beginning. The chatbot is set to evolve into a personal assistant, with Jony Ive’s io (now part of OpenAI) producing the specific hardware.
OpenAI’s Sam Altman has repeatedly hinted at the future of ChatGPT, an AI that comprehends everything about you and can assist in numerous ways. The leaked information regarding the ChatGPT io device also aligns with this vision. The device will be positioned on desks, worn around our necks, or kept in our pockets, equipped with new social and ambient computing capabilities that enable it to function as an always-active assistant able to distinguish between different speakers.
OpenAI has not yet unveiled the device or the ChatGPT functionalities that would enable the chatbot to act as an assistant. For this to occur, the AI must access various applications on both mobile phones and computers, such as email, messaging, and calendars, along with mastering social nuances as part of the ambient computing layer being developed by OpenAI.
Nevertheless, the AI organization has now confirmed its intentions for a “super-assistant” through documentation submitted to the court in the Google antitrust lawsuit. These plans are notably ambitious as well. This is anticipated to take place this year, well ahead of the official announcement for the ChatGPT io hardware device.
The legal document (as reported by The Verge) titled “ChatGPT: H1 2025 Strategy” is extensively redacted, likely to hide certain information that OpenAI does not want disclosed to the public before an official launch. The document was drafted last year, as evidenced by its discussion of the two segments of 2025 in the context of OpenAI’s product development strategy.
However, several paragraphs remain unredacted in the document, clearly outlining OpenAI’s goals for ChatGPT this year:
In the first half of next year, we plan to transform ChatGPT into a super-assistant: one that understands you, knows what matters to you, and supports any task that a smart, reliable, emotionally intelligent individual with a computer could perform. The timing is appropriate. Models like o2 and o3 have finally reached a level of intelligence that allows them to consistently carry out agentic tasks, tools like computer usage will enhance ChatGPT’s capacity to act, and interaction methods such as multimodality and generative UI permit both ChatGPT and users to communicate in the most effective manner for the task at hand.
What does a super-assistant entail? It is an intelligent entity with T-shaped competencies. It’s considered an entity because it is customized to you and accessible wherever you are, including chatgpt.com, our native applications, telephone, email, or third-party platforms like Siri. It’s T-shaped because it possesses comprehensive skills for routine tasks that may seem tedious, along with specialized expertise for tasks that many individuals find daunting (starting with coding). The broad aspect focuses on simplifying life: answering inquiries, locating homes, getting in touch with lawyers, signing up for gyms, organizing trips, purchasing gifts, managing schedules, and keeping tabs on tasks, while the deep aspect concentrates on
As a long-time ChatGPT user, I can affirm that ChatGPT is not a super-assistant at the moment. It cannot be considered as such since it lacks unfettered access to user data, particularly on mobile devices. Even if it had such access, I would hesitate to place my trust in an AI model granted access to personal data without assurance of how that data is safeguarded and without explicit privacy measures in place.
